AI-native systems for code intelligence, agentic research, and legal technology.
AI Indie Developer · Product Manager · Shanghai, China
I turn knowledge-heavy workflows into open-source products: systems that explain codebases, coordinate research agents, and make domain expertise operational.
- Code → knowledge. Bilingual documentation, code graphs, and explainable retrieval.
- Signals → evidence. Multi-agent research, forecasting, and report generation.
- Expertise → workflow. Domain-specific AI for legal work and long-form creation.
 Python |
 TypeScript |
 Rust |
 Next.js |
 Node.js |
 FastAPI |
 PostgreSQL |
 Docker |
| 01 OpenWiki |
Turns repositories into explorable bilingual wikis, code graphs, and RAG-powered research spaces.Rust Next.js RAG · Original project |
| 02 OpenFish |
Coordinates agents for cross-platform public-opinion research, forecasting, and report generation.Python Multi-agent Research · Original project |
| 03 InkFlow |
An AI copilot for long-form fiction, world-building, and story consistency.TypeScript GraphRAG Knowledge graph · Original project |
| 04 Claude for Legal · 中国法 |
Adapts legal agents, skills, and MCP connectors for Chinese legal practice in Claude Code and Codex.Python MCP Legal AI · Upstream adaptation |
- Building reliable agentic systems that move from research to a usable result.
- Exploring code intelligence, explainable retrieval, and local-first knowledge systems.
- Translating legal and creative expertise into inspectable AI workflows.
- Original projects — software designed and built here.
- Customized forks — upstream projects adapted for a specific domain or workflow.
- Reference forks — research material kept for comparison and learning.
- AI agents · Legal technology — browse by technical focus.
Selected reference repositories
- BettaFish — multi-agent public-opinion analysis.
- hermes-agent — adaptive personal agents.
- MediaCrawler — multi-platform social-media collection.
我是 Weiyang，一名 AI 独立开发者与产品经理。关注 AI Agent、代码智能、法律科技与复杂知识工作流，持续把研究型原型打磨成可理解、可检查、可使用的开源产品。